E H Chamberlin: Oligopoly, and Oligopolistic Interdependence: The Issue of Space I INTRODUCTION The late 1920s and early 1930s saw considerable activity amongst economists concerned with competitive structures and the "firm". Much of this work may be interpreted as an attack on Marshall's treatment of the subject with a view to replacing it by a more "rigorous" and formal analysis. But E H Chamberlin to a very large extent stands apart from these developments, as he makes plain in the "Origin and Early Development of Monopolistic Competition Theory" (1961). Serious work on his thesis apparently began in 1924, was largely completed in 1926, and the study filed in the following year. This means, for example, that Chamberlin's "discovery" of the curves of marginal cost and marginal revenue was made quite independently of his English and German colleagues. Further, as Chamberlin himself made clear, "Nor did the Book itself attack Marshall...on any of the issues involved" (ibid, p 532). Indeed, he always insisted that his work was an attack "not on Marshall but on the theory of perfect competition" (ibid, p 540). He might have added that Monopolistic Competition is essentially Marshallian both in its style of reasoning and in the pre-occupation with realism; a pre-occupation which led Chamberlin to play down the operational significance of the marginal curves while recognising their importance in a technical sense (1957, p 274-6). In contrast to the origins of Joan Robinson's Imperfect Competition, Chamberlin cited three empirical sources of inspiration. First, the "Taussig-Pigou controversy as to whether charging what the market will bear in railway rates was to be explained in terms of monopolistic discrimination or in terms of joint costs" (1961, p 517). Secondly, he drew attention to the "literature of business" and especially to the stimulus provided by Allyn Young. As Chamberlin recalled, Young had given particular attention to trade marks and 1 patents in so far as they conferred a monopoly power which is also consistent with competition: "Each makes a product unique in certain respects; this is its monopolistic aspect. Each leaves room for other commodities almost but not quite like it; this is its competitive aspect" (quoted by Chamberlin, ibid, p 525). For Chamberlin, this perspective "became the key to the whole analysis" (ibid, p 526); a new "way of looking at the economic process" (1948, p 204). A third source of stimulus was located in "business economics" with special reference to advertising and the operation of retail markets (1961, p 529). Here Chamberlin drew attention to Pigou's perception that advertising was peculiar to what was in effect monopolistic competition, and to Marshall's distinction between "constructive" and "combative" advertising. II Chamberlin's main interest was in that situation where the pairing of buyers and sellers was no longer random: that is, where goods are differentiated in the mind of the consumer. Here the firm may control price, product specification and selling costs where the latter are "incurred to alter the position or shape of the demand curve" (1948, p 117). The problems which are exposed by such a perspective are wide ranging so that it is important to note that Chamberlin concerned himself at the outset almost solely with intra rather inter-industry competition and that he retained something very like the Marshallian industry in so doing. As he put it: 2 "The group contemplated initially is one which has ordinarily been regarded as composing one imperfectly competitive market: a number of automobile manufacturers, or producers of pots and pans, magazine publishers, or of retail shoe dealers" (ibid, p 81). Even this restricted perspective, however, meant that the technical apparatus had to be altered as compared to the treatment of pure competition. The firm now has a choice of cost curve arising from the capacity to alter product specification (ibid, pp 94-100) while in addition selling costs have to be added to costs of production. Chamberlin thought that it would be difficult to generalise about the curve of selling costs since its shape would be affected by the nature of the product, the level of planned expenditure, and the choice of media, but gave it a "U" shape in both short and long run (ibid, pp 133-5, 138). Perhaps the most important innovation resulted from Chamberlin's appreciation of the point that: "A monopoly of Lucky Strikes does not constitute a monopoly of cigarettes, for there is no degree of control whatever over the supply of other substitute brands" (ibid, p 65). He thus introduced two revenue curves. The DD curve is defined as the fractional part of the Market Demand curve (eg, for cigarettes) and "shows the demand for the product of any one seller at various prices on the assumption that his competitors' prices are always identical with his" (ibid, p 90). In contrast, the dd curve provides "a rough index of buyers' preferences for the product of one seller over that of another" (ibid, p 93) and indicates the "increase in sales which he could realise by cutting his price provided others did not also cut theirs" (ibid, p 90). In the context of the large group, the relatively elastic (as compared to DD) dd curve is relevant since it is legitimate to assume that any adjustment "by a single producer spreads its 3 influence over so many of his competitors that the impact felt by any one of them is negligible" (ibid, p 83) – ie the diffusion effect. To ease the task of exposition, Chamberlin proceeded under the "heroic" assumption that both demand and cost curves for all products " are uniform throughout the group" (ibid, p 82) thus permitting the use of a single set of revenue and cost functions - and further heightening the misleading impression that the reader confronts a minor variation of the earlier treatment of pure competition. But methodologically, the technique is classically Marshallian in that Chamberlin assumed price and product, product and selling costs, selling costs and price to be given, before allowing the third element in each case to vary (ibid, chapters 5 and 7). In fact, Chamberlin spent a great deal of time in discussing product variation and selling costs although there is little doubt that the analysis of price adjustment attracted most attention from later commentators. But the procedure is familiar: if the typical seller faces a dd curve which is consistent with excess profits, new entry will force the curve to a position of tangency with the cost curve (PP) thus yielding an equilibrium under conditions of decreasing costs – a variant of the Sraffa solution (1953). While Chamberlin makes allowance for changes in the position of the cost curve in consequence of new entry, an interesting example is where numbers are assumed constant and excess profits obtain. In this case, the (unintended) consequences for each seller seeking to cut price with a view to maximising profit, is a general reduction in the market price so that dd "slides" down DD until the position of tangency is reached. Equilibrium is defined by two conditions: "(a) dd' must be tangent to PP', and (b) DD' must intersect both dd' and PP' at the point of tangency" (ibid, p 93). 4 However, the emphasis on tangency obscures the real purpose, and one must sympathise with Chamberlin's repeated complaint that this solution had come to be regarded as "the central principle involved" (ibid, p 195). In fact, Chamberlin recognised that decisions on cost, price, and selling outlay were closely linked, and therefore that the cost and revenue functions facing the firm were interdependent. The reader is, in fact, presented with a market situation where change is endogenous, in part the result of firms seeking positions of advantage for themselves, so that in practice the "result is heterogeneity of prices, and variation over a wide range of outputs (scales of production) and in profits" (ibid, p 81) which are unlikely to be completely eliminated even in a competitive environment. Figure 1 (1948, p91) III DUOPOLY However, to the modern reader as for Chamberlin himself, the third chapter on "Duopoly and Oligopoly" (which also figured in the thesis, pp 63-96) should be among the most interesting. The material was first published in the Quarterly Journal for 1929. Chamberlin was obliged 5 to strike out the reference to "oligopoly" at the insistence of F W Taussig, the then editor, who "thought the term a monstrosity" (1957, p33); a decision which must have cost the author some pain since he was unaware at that time of any prior use of the term in print. Chamberlin fared better at the hands of later generations. Romney Robinson, for example, insisted that the introduction of the analysis of oligopoly was to be regarded as the most important contribution of a writer who "almost single-handedly introduced the concept …and forced it to the attention of economists" (1971, p 63). Bain concurred (1964, p 29). It would be inappropriate here to review the extensive history of duopoly theory or even the treatment which Chamberlin accorded to the subject in his third chapter (and Appendix A which deals with "Mathematical Theories of Duopoly and Oligopoly"). However, it is appropriate to note that the Cournot model (1838) of two sellers offering an identical product is arguably the most important at least for didactic purposes. Basically, Chamberlin reviewed a case where the sellers were interdependent but pursued policies with regard to price or output on the assumption that their rival's reaction would remain unchanged - the case of mutual dependence ignored. In both cases (price and output), Chamberlin confirmed that the outcome would be determinate and accord with the situation which would have obtained under competitive conditions.